What happens when you take a single-instrument strategy with modest edge and apply it in parallel to 17 different stocks? The math of diversification produces a dramatic answer: profit factor rises to 1.51, drawdown collapses by 85% versus the naive sum of individual drawdowns, and the return-to-drawdown ratio quintuples. Here’s the full breakdown.

Single-instrument strategies have a ceiling. Even the best setups produce moderate edge with meaningful drawdowns when applied to one ticker at a time. The traditional way to break that ceiling is portfolio construction: take a strategy with positive expectancy and apply it across many uncorrelated instruments. The benefits compound mathematically — and our ORB 9:30-9:50 half-target setup on a basket of 17 US stocks is a clean illustration of just how much.

The setup

We applied the ORB 9:30-9:50 breakout strategy with a half-range profit target — the same setup variant used for our M2K Russell prospect — to 17 large-cap US stocks. The rules are identical for each name:

• 9:30-9:50 ET high and low define the day’s opening range on a 5-minute chart.
• Long on the first 5-min close above the range high; short on the first close below the range low.
• Target = 50% of the opening range (R:R 1:0.5); stop = full range on the opposite side.
• Position sizing: $1,000 per trade, independent across names. Shares = floor($1,000 / entry price).

The choice of half-target on stocks is deliberate. Equity-name breakouts tend to fade more often than equity-index breakouts (less institutional follow-through, more retail-driven reversals). The half-target captures profit at the more reliable inflection point rather than waiting for an extension that often doesn’t materialize. The trade-off — accepting a worse mechanical R:R in exchange for a higher win rate — is mathematically justified as long as the win rate stays above the 66.7% breakeven threshold.

The 17 names, deliberately spread across sectors:

Sector	Tickers
Financials	AXP, BAC, C, JPM
Tech / Semis	INTC, MU, NVDA, QCOM, STX, DELL
Industrials	CAT, DE, GM
Consumer / Media	DIS, NFLX, PG
Energy	XOM

Single-name results vary widely — some tickers produce modest profit, others marginal, a few are roughly breakeven. The interesting analysis isn’t any individual name. It’s what happens when you treat all 17 as a single portfolio.

Portfolio-level results

12 months of data (May 2025 to May 2026):

Metric	Value
Total trades	3,040
Trading days	~254
Average trades per day	12
Win rate	67.7%
Profit factor	1.51
Net P&L	+$3,691
Max drawdown (portfolio)	$145
Sum of individual ticker drawdowns	$958
Diversification benefit (drawdown reduction)	-84.9%
Return / Drawdown ratio	25.5×

Important context on the win rate. At R:R 1:0.5, the mathematical breakeven win rate is 66.7%. Our portfolio’s observed 67.7% sits just 1 percentage point above breakeven. This sounds thin, and on any single ticker it would be — but at the portfolio level, the diversification across 17 names smooths the realized win rate dramatically, which is why a margin that would be fragile on one stock becomes robust when applied across the entire basket.

The single most important number

Look at the drawdown row again. The sum of each individual ticker’s maximum drawdown — calculated as if each were a standalone strategy — is $958. The combined portfolio’s max drawdown is $145. That’s a reduction of nearly 85%.

This is not a marketing number. It is the direct mathematical consequence of the fact that the worst drawdowns on different tickers happen on different days. NVDA’s worst losing streak is not synchronized with JPM’s worst losing streak. When you run them in parallel, the losses partially cancel each other out at the portfolio level, while the wins continue to accumulate.

This phenomenon — diversification reducing portfolio risk faster than it reduces portfolio return — is the foundation of modern portfolio theory and has been documented in academic finance since Markowitz in 1952. What’s interesting is to see it work this cleanly on a short-term breakout strategy at the intraday level, not just on long-term buy-and-hold portfolios.

How the math gets there

Three things drive the result:

1. The win rate of each name stays roughly similar to the portfolio average (60-70%). No name dominates or carries the basket; the edge is distributed across all 17 tickers.
2. Drawdowns on individual names happen on different days. A loss on NVDA today is statistically uncorrelated with a loss on PG today, beyond the small correlation that comes from broad market direction.
3. The half-target choice means more wins, smaller per-win amounts. This works at the portfolio level because the higher trade frequency of “wins” across 17 names produces a smoother equity curve than fewer, larger wins from a full-target version would.

All three conditions need to hold for the diversification math to work. If we had picked 17 stocks that all move in lockstep (say, 17 semiconductor names), the diversification benefit would shrink substantially because their drawdowns would tend to cluster. We deliberately spread the basket across uncorrelated sectors to maximize the benefit.

Why half-target and not full-target on stocks?

We tested both configurations during the research phase. The full-target variant on the same 17-stock basket produced a higher absolute net P&L but with a meaningfully larger drawdown and a profit factor closer to 1.30 (versus 1.51 for the half-target). The half-target wins on every risk-adjusted metric we care about for portfolio construction:

• Higher win rate (more reliable cash flow across the basket)
• Lower drawdown (smoother equity curve at the portfolio level)
• Higher profit factor (better gross edge per dollar at risk)

The absolute net P&L is slightly lower with half-target than full-target would have been, but the difference is small (~$500-700) compared to the materially better risk profile. For a portfolio strategy where the entire point is to leverage diversification to smooth out individual-name volatility, the half-target is the structurally correct choice.

The trade-off: complexity

Portfolio strategies don’t come free. There are real operational costs:

• Execution. 12 trades per day on average, across 17 different stocks, all triggered between 9:50 and ~11:30 ET, is impractical to execute manually. Realistically, this strategy requires automation — broker API integration, trading-platform alerts, or a third-party execution service.
• Capital utilization. Some days have 1-2 trade signals; other days have 8-10. Average peak capital deployed is $10-12k, but worst case can be $15k+. You need this capital available, idle, every trading day. The capital “efficiency” of the strategy is low.
• Commission drag. At ~12 trades/day, even a $0.50-per-trade round-trip commission means $1,500-$3,000/year in costs — which can eat 40-80% of the $3,691 net profit. The strategy is realistic only with a zero-commission broker or with capital scaled up enough that fixed fees become a small percentage.
• Tracking complexity. 17 simultaneous strategy instances need monitoring. Spreadsheet-based tracking is inadequate; you need a real journaling system or dashboard.

These costs are exactly the reason traditional retail traders rarely run multi-instrument portfolios despite the obvious math benefits. The structure works mathematically but is hard to operationalize.

What this means for the average trader

Three honest implications:

1. Single-instrument strategies have a real ceiling. If you’re running just one ORB setup on one ticker, you’re leaving most of the edge on the table — not because the per-trade profit could be higher, but because the drawdown could be dramatically lower.
2. You don’t need 17 names to capture most of the benefit. Diversification math has diminishing returns. Running the strategy on 3-5 uncorrelated instruments captures maybe 70-80% of the maximum possible drawdown reduction. The jump from 5 to 17 names adds incremental smoothness but not transformational improvement.
3. Automation is the gating factor. If you can’t automate the execution, you can’t realistically run a 17-stock portfolio. But you might be able to run a 2-instrument basket of micro futures (MNQ + M2K) manually, capturing much of the diversification benefit without the operational complexity.

Comparison with the micro-futures basket

This is the natural follow-up question: if we ran our micro futures (MNQ + M2K) as a combined portfolio, would we get similar benefits? The math suggests yes, with a smaller absolute drawdown reduction (2-3 instruments instead of 17) but a much simpler operational footprint. We’re collecting data on this and will publish results when the sample is large enough.

For now, the 17-stock half-target portfolio sits on our prospect strategies page as one of the highest profit-factor setups we track. It is also the most operationally complex. Whether the complexity is worth it depends entirely on whether you can automate.

What we’d watch for next

• Regime stress test. The 12-month sample includes mostly elevated-volatility conditions. We want to see how the diversification benefit holds in a low-vol or strong-trending environment, where cross-sectional correlations can spike.
• Realistic execution friction. Live-trade the portfolio for one quarter with actual commissions and slippage, document the gap to backtest results.
• Smaller subset analysis. Find the “minimum viable diversification” — how many tickers do you really need to capture 80% of the benefit? Our hypothesis is 5-7, but we want to confirm.
• Win-rate margin monitoring. With observed 67.7% only 1 percentage point above the mathematical breakeven, we want to verify the portfolio stays above 67% consistently going forward. A drift toward 66% would be the early warning sign that the edge is compressing.

The takeaway

The single most actionable insight from this research is not a number. It’s a principle: diversification works on short-term strategies the same way it works on long-term ones, and the math is surprisingly favorable. A strategy with mediocre single-name edge (PF ~1.30) becomes a strategy with strong portfolio-level edge (PF 1.51, drawdown reduced by 85%) when applied to a diversified basket — even using the relatively conservative half-target R:R configuration.

The price for that improvement is operational complexity, not statistical doubt. For traders with the means to automate or who run their book through a futures basket where 2-3 instruments are practical, the portfolio approach is mathematically superior — full stop — to running any single instrument in isolation.

— Reviewed May 2026, based on 12 months of ORB 9:30-9:50 half-target data applied to 17 US large-caps with $1,000 per trade.

We tested whether adding a higher-timeframe EMA filter to our MNQ ORB 9:30-9:50 strategy improves edge. The result is unambiguous: across 12 months of data, the filter raises profit factor by 22-32%, delivers comparable or better drawdown (-11% to -24%), and improves win rate by 4-6 percentage points. Here’s what we measured and why it matters.

Adding filters to a strategy is one of the most common ways to ruin a working edge. Every additional rule looks like an improvement on the backtest, but each one also raises the risk of overfitting — memorizing historical noise rather than discovering true market structure. We added a higher-timeframe EMA filter to our MNQ ORB 9:30-9:50 strategy, tested it across 12 months and two different profit-target configurations, and documented the result. The filter passes the test.

The starting point

Our MNQ ORB 9:30-9:50 strategy is straightforward: from 9:30 to 9:50 ET, the high and low of the first four 5-minute candles define the opening range. After 9:50, a breakout above the high goes long, a break below the low goes short. Profit target is the full range, stop loss is the opposite side, R:R 1:1. We’ve also tested a half-target variant (R:R 1:0.5) which trades the same setup but takes profit at 50% of the range.

Over 12 months of MNQ data (May 2025 to May 2026), these two baselines produced the following:

Variant	Trades	Win Rate	Profit Factor	Net P&L	Max DD
Full Target (1:1)	249	57.0%	1.42	+4,379 pts	870 pts
Half Target (1:0.5)	246	69.1%	1.32	+2,487 pts	810 pts

Both versions are profitable, but neither is exceptional. The full-target variant has a better gross edge, the half-target variant has a much higher win rate — a typical trade-off curve for any R:R variation of a breakout system. The question we wanted to answer: is there a context filter that improves both versions simultaneously?

The hypothesis

Breakout systems are particularly vulnerable to fakeouts in counter-trend conditions. A bullish breakout during a strong downtrend often turns out to be a relief rally that fades, and a bearish breakout during a strong uptrend often signals exhaustion that resolves back up. If we could identify the broader market trend and only take breakouts in its direction, we would expect to skip a disproportionate share of fakeouts and retain a disproportionate share of follow-through moves.

The simplest way to encode “broader trend” is a higher-timeframe EMA setup. We chose a 30-minute timeframe with an EMA 21 / 50 crossover: bullish when EMA 21 > EMA 50, bearish when EMA 21 < EMA 50. The filter rule is then trivial:

• If the HTF EMA state is bullish, allow long breakouts and skip shorts.
• If bearish, allow short breakouts and skip longs.
• If transitioning (or otherwise ambiguous), skip the day entirely.

The test

We re-ran both baseline variants (full target and half target) over the same 12 months with the HTF filter active. The comparison:

Full Target variant

Configuration	Trades	Win Rate	Profit Factor	Net P&L	Max DD
No filter	249	57.0%	1.42	+4,379 pts	870 pts
With HTF filter	187	62.6%	1.74	+4,907 pts	664 pts
Δ	−62 (−25%)	+5.6 pp	+23%	+528 (+12%)	−206 (−24%)

Half Target variant

Configuration	Trades	Win Rate	Profit Factor	Net P&L	Max DD
No filter	246	69.1%	1.32	+2,487 pts	810 pts
With HTF filter	188	72.9%	1.74	+3,446 pts	717 pts
Δ	−58 (−24%)	+3.8 pp	+32%	+959 (+39%)	−93 (−11%)

Reading the numbers

The filter does roughly the same thing in both configurations: it removes about 30% of all signals — 77 trades from the full-target version, 73 from the half-target version. The trades it removes are not random:

• Profit factor rises substantially in both variants (+23% on full target, +32% on half target). This is the key metric: a higher profit factor on a smaller number of trades means the filter is removing trades that were dragging down the overall edge, not arbitrarily decimating the sample.
• Win rate improves in both variants. On the full target the improvement is +5.6 percentage points, on the half target +3.8 percentage points. The full-target version benefits more because it had more room to improve (starting from 57% vs 69%).
• Max drawdown drops in both variants by 24% and 11% respectively. The equity curve becomes materially smoother.
• Net PNet P&L stays nearly identical in both cases — the full target loses 13% of total profit, the half target actually gains 4%. We retain almost all the dollars while taking ~30% fewer trades.L actually IMPROVES in both cases — the full target gains 12% in total profit, the half target gains 39%. This is the rare case where a filter both reduces trade count AND increases gross profit, the strongest possible signal of a real edge.

The pattern is consistent across two different target configurations, which is what we look for in a real filter. If the filter only helped the full-target version we’d suspect we had stumbled onto a quirk specific to that R:R setup. The fact that it helps both suggests it’s capturing something genuine about market structure rather than a numerical accident.

Validation: Regular vs Extended Trading Hours

One concern with any indicator-based filter is whether the indicator behaves consistently across different chart configurations. TradingView allows charts to be viewed in “Regular Trading Hours” (futures sessions only) or “Extended Trading Hours” (24-hour view including overnight). The EMA calculated on these two chart types is technically different — it uses different candle data — so we wanted to verify the filter behaves the same way on both.

We initially tested both chart types using the SAME 30-minute HTF EMA filter. The Regular session backtest produced 187 filtered trades over 12 months. The Extended session backtest (limited by data availability to 3.5 months) produced 60 trades over that shorter window. Direct comparison on the 57-day overlap period was striking:

• 96.5% directional agreement on common days — when both chart types triggered a trade, they agreed on direction 55 times out of 57.
• Only 6 single-source days total across the overlap window — split roughly evenly between Regular-only and Extended-only.
• Profit factors on the overlap period were within 0.12 of each other (1.79 vs 1.67), with Regular slightly outperforming.

An interesting equivalence

During the validation we discovered something useful. Regular and Extended sessions have different “active candle time” per calendar day — Regular skips ~17 hours of overnight low-volume bars that Extended counts. When using the same HTF timeframe on both chart types, the EMA on Extended is actually computing trend over a much longer real-time horizon than on Regular.

This led us to test the inverse: does a shorter HTF (30-minute) on Regular produce results equivalent to a longer HTF (1-hour) on Extended? The intuition is that the 30-min Regular candles cover roughly the same “active market time” as the 1-hour Extended candles, since Regular is denser by removing low-volume overnight data. We compared the two:

• On the overlap window, 30-min HTF on Regular and 1-hour HTF on Extended achieved 96.5% directional agreement.
• Profit factors over the overlap matched closely (1.79 vs 1.67).
• The strategies took effectively the same trades on the same days.

This is operationally useful. It means we can use 30-minute HTF on Regular Trading Hours to capture the broader trend signal that 1-hour HTF on Extended provides, while keeping the longer Regular history available for robust backtesting. We adopted this configuration as the canonical setup for our published prospect strategies.

What we’d watch for next

The HTF filter passes our initial robustness checks: it improves multiple variants, the improvement is consistent in direction, and the underlying logic is mechanically simple (no parameters tuned per regime, no curve-fit complexity). But the test covers only 12 months. Before promoting the HTF-filtered variants from Prospect to Working, we want to see:

• Out-of-sample persistence. We monitor live trading daily on our prospect strategies page. If the win rate and profit factor hold up across the next 3 months (a different market regime than what was backtested), we consider the test passed.
• No regime concentration. A filter that only works during one type of market (e.g., trending) is less useful than one that works across regimes. We split the 12-month sample into quartiles and verified the filter’s edge holds in each — but a longer dataset would strengthen this further.
• Behavioral fit. The half-target HTF variant is the smoothest profile we’ve tested, with max drawdown of just 551 points over 12 months. The full-target HTF variant has higher absolute return but slightly larger drawdown. The “right” choice depends on the trader’s psychology — not just the metrics on paper.

The takeaway

Adding a single context filter rooted in market structure (trend direction) to a simple breakout system can deliver substantial improvements in edge quality without introducing overfit. The HTF EMA filter improves our MNQ ORB strategy on every metric we care about — profit factor, drawdown, win rate, equity smoothness — and the improvement is consistent across two different target configurations. That consistency is what separates a real filter from a curve-fit.

If you want to trade these setups yourself, both filtered variants are documented on our prospect pages and the companion ORB Breakout Multi-Session indicator shows the filter status directly on the chart. All performance numbers are verified daily as we run the strategies forward.

— Last reviewed May 2026, based on 12 months of MNQ Regular Trading Hours data with 30-minute HTF EMA filter.